On Sat, Jun 28, 2025, ZIHAO BAI via AMBER wrote:
One additional point:
>
>In my test, I simulated a system of 489 methanol molecules using a
>classical MM force field. I then converted one methanol molecule to the QM
>region using DFTB3 (via DFTB+), keeping the rest as MM.
I'm not sure this this is/was just a test, but I'll make a general comment
about QM/MM in Amber: The code allows you to choose almost any subset of
atoms as the QM region. But this doesn't mean the everything will work
equally well.
The original design was for a single, compact QM region, that was large
enough for the boundary between QM and MM to be a ways away from the
"important" part of the QM region. Even if no covalent bonds are broken,
the boundary is always a bit approximate: it relies upon artificial
Lennard-Jones parameters assigned to the QM atoms to keep QM and MM atoms
apart from one another. Further, point charges in the MM region interact
with the QM Hamiltonian in ways that only approximate what a "true" QM-QM
electrostatic region would describe, especially at short distances.
So, having a QM region with only 6 atoms is a bit outside the original
design parameters. Others on the list might want to chime in with their
experience and advice here.
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Jun 28 2025 - 16:00:03 PDT
